1BUNDLE-EXEC(1) BUNDLE-EXEC(1)
2
6 bundle-exec - Execute a command in the context of the bundle
7
9 bundle exec [--keep-file-descriptors] command
10
12 This command executes the command, making all gems specified in the
13 [Gemfile(5)][Gemfile(5)] available to require in Ruby programs.
14 Essentially, if you would normally have run something like rspec
16 spec/my_spec.rb, and you want to use the gems specified in the [Gem‐
17 file(5)][Gemfile(5)] and installed via bundle install(1) bundle-in‐
18 stall.1.html, you should run bundle exec rspec spec/my_spec.rb.
19 Note that bundle exec does not require that an executable is available
21 on your shell´s $PATH.
22
24 --keep-file-descriptors
25 Exec in Ruby 2.0 began discarding non-standard file descriptors.
26 When this flag is passed, exec will revert to the 1.9 behaviour
27 of passing all file descriptors to the new process.
28
30 If you use the --binstubs flag in bundle install(1) bundle-in‐
31 stall.1.html, Bundler will automatically create a directory (which de‐
32 faults to app_root/bin) containing all of the executables available
33 from gems in the bundle.
34 After using --binstubs, bin/rspec spec/my_spec.rb is identical to bun‐
36 dle exec rspec spec/my_spec.rb.
37
39 bundle exec makes a number of changes to the shell environment, then
40 executes the command you specify in full.
41 • make sure that it´s still possible to shell out to bundle from in‐
43 side a command invoked by bundle exec (using $BUNDLE_BIN_PATH)
44 • put the directory containing executables (like rails, rspec,
46 rackup) for your bundle on $PATH
47 • make sure that if bundler is invoked in the subshell, it uses the
49 same Gemfile (by setting BUNDLE_GEMFILE)
50 • add -rbundler/setup to $RUBYOPT, which makes sure that Ruby pro‐
52 grams invoked in the subshell can see the gems in the bundle
53 It also modifies Rubygems:
57 • disallow loading additional gems not in the bundle
59 • modify the gem method to be a no-op if a gem matching the require‐
61 ments is in the bundle, and to raise a Gem::LoadError if it´s not
62 • Define Gem.refresh to be a no-op, since the source index is always
64 frozen when using bundler, and to prevent gems from the system
65 leaking into the environment
66 • Override Gem.bin_path to use the gems in the bundle, making system
68 executables work
69 • Add all gems in the bundle into Gem.loaded_specs
71 Finally, bundle exec also implicitly modifies Gemfile.lock if the lock‐
75 file and the Gemfile do not match. Bundler needs the Gemfile to deter‐
76 mine things such as a gem´s groups, autorequire, and platforms, etc.,
77 and that information isn´t stored in the lockfile. The Gemfile and
78 lockfile must be synced in order to bundle exec successfully, so bundle
79 exec updates the lockfile beforehand.
80 Loading
82 By default, when attempting to bundle exec to a file with a ruby she‐
83 bang, Bundler will Kernel.load that file instead of using Kernel.exec.
84 For the vast majority of cases, this is a performance improvement. In a
85 rare few cases, this could cause some subtle side-effects (such as de‐
86 pendence on the exact contents of $0 or __FILE__) and the optimization
87 can be disabled by enabling the disable_exec_load setting.
88 Shelling out
90 Any Ruby code that opens a subshell (like system, backticks, or %x{})
91 will automatically use the current Bundler environment. If you need to
92 shell out to a Ruby command that is not part of your current bundle,
93 use the with_unbundled_env method with a block. Any subshells created
94 inside the block will be given the environment present before Bundler
95 was activated. For example, Homebrew commands run Ruby, but don´t work
96 inside a bundle:
97 Bundler.with_unbundled_env do
101 `brew install wget`
102 end
103 Using with_unbundled_env is also necessary if you are shelling out to a
107 different bundle. Any Bundler commands run in a subshell will inherit
108 the current Gemfile, so commands that need to run in the context of a
109 different bundle also need to use with_unbundled_env.
110 Bundler.with_unbundled_env do
114 Dir.chdir "/other/bundler/project" do
115 `bundle exec ./script`
116 end
117 end
118 Bundler provides convenience helpers that wrap system and exec, and
122 they can be used like this:
123 Bundler.clean_system(´brew install wget´)
127 Bundler.clean_exec(´brew install wget´)
128
132 At present, the Rubygems plugin system requires all files named
133 rubygems_plugin.rb on the load path of any installed gem when any Ruby
134 code requires rubygems.rb. This includes executables installed into the
135 system, like rails, rackup, and rspec.
136 Since Rubygems plugins can contain arbitrary Ruby code, they commonly
138 end up activating themselves or their dependencies.
139 For instance, the gemcutter 0.5 gem depended on json_pure. If you had
141 that version of gemcutter installed (even if you also had a newer ver‐
142 sion without this problem), Rubygems would activate gemcutter 0.5 and
143 json_pure <latest>.
144 If your Gemfile(5) also contained json_pure (or a gem with a dependency
146 on json_pure), the latest version on your system might conflict with
147 the version in your Gemfile(5), or the snapshot version in your Gem‐
148 file.lock.
149 If this happens, bundler will say:
151 You have already activated json_pure 1.4.6 but your Gemfile
155 requires json_pure 1.4.3. Consider using bundle exec.
156 In this situation, you almost certainly want to remove the underlying
160 gem with the problematic gem plugin. In general, the authors of these
161 plugins (in this case, the gemcutter gem) have released newer versions
162 that are more careful in their plugins.
163 You can find a list of all the gems containing gem plugins by running
165 ruby -e "puts Gem.find_files(´rubygems_plugin.rb´)"
169 At the very least, you should remove all but the newest version of each
173 gem plugin, and also remove all gem plugins that you aren´t using (gem
174 uninstall gem_name).
175 February 2023 BUNDLE-EXEC(1)